Detecting Pesticides in UIUC Greenhouses Save the Bees Nat, Dav, - - PowerPoint PPT Presentation

Detecting Pesticides in UIUC Greenhouses Save the Bees

Nat, Dav, Mat, Zak

Introduction

• Purpose - acquire more data on pesticide use

in order to learn more about their potential externalities

• Particulate Sensor measures how long

aqueous pesticide particles persist in the air

• American study shows the rusty-patched

bumble bee and monarch butterfly populations have decreased by ~90%

Hardware

• Arduino
• BME680 (temp., humidity, gas sensor, pressure)
• PM2.5 Air Quality Sensor
• Ultimate GPS Breakout
• Precision I2C Real Time Clock
• MicroSD Breakout
• INA219 DC Current Sensor
• Keypad
• Liquid Crystal Display screen

Hardware - PM2.5 Air Quality Sensor

PM2.5 Air Quality Sensor was the most important sensor used in our experiment since we were interested in determining the air quality, or persistence of pesticide particles in the air over a range of time.

• Uses light scattering to measure the

amount of suspended particles in the air

• Collects data for 0.1L of air at a time

with the fan inside the device

• Sensor specifications show that it data

collection was very precise in the temperature of the greenhouses

Data Acquisition

• DAQ loop ran 20x per second
• Opened SD card on initial run
• SD file remained open until one of:
• Batteries reaching critical level
• Specified time limit reached
• Manually turned off using keypad

Data Acquisition

• Data printed to SD card included:
• BME680 Temperature, Pressure, Humidity, and Altitude
• Particulate data of various sizes
• Voltage, current, and power
• GPS information when relevant
• Datum separated by commas
• Headers were printed at top

Data Acquisition Testing

• Testing of device was done on various sources:
• Soldering station
• Grilled cheese sandwich
• Flour in air
• Sawing
• Walking around Loomis

Experiment

• Experiments were originally planned

to be done outside on fields. However, due to the fact that spraying happens in June and July, we had to conduct our experiments in green houses (30 ft by 15 ft, with a 12 ft ceiling)

• UIUC Green houses were sprayed by

a licensed professional (S/O to Monty)

Experiment - Pesticides and Spraying details

• Various aqueous pesticide solutions

were used in the spraying process

• Some were known to be more

harmful to the environment and people’s health than others

• Ex. Buprofezin is labeled as a health

and and environmental hazard

• Spraying occured around 4pm
• Re-entry was not allowed until 7am

the following day

Experiment - Results

• Spraying began at the

start of the increased particle amounts.

• After the spraying, the

particulate count falls exponentially Data Analysis Details Explained in Next Slides

Experiment - Biggest Challenge to Overcome

Battery Life and Power

• 10 1.5 Volt batteries were used in
• ur device after seeing that 5

batteries did not last long enough

• The device with 5 batteries only

lasted approximitly 4 hours.

• Devices with 10 batteries were able

to last more than double that time *Blue: 5, Red 10 batteries

Experiment - Other Data Collection Challenges

• Files would get deleted when the battery pack died. In order to avoid this,

code was added to have the device save the data once the total battery voltage reached 6.7V

• If the device was still running when we got there, we close the file by

pressing a button on the keypad (#) then powered off the device

• Files would not be created with a large amount of characters (This one

took a while to figure out)

Data Analysis - Cleanup

• Files directly from SD card needed clean up
• Missing data from first few seconds
• Shifted rows
• Bad time values
• Files first opened using Excel
• Data was manually searched for error
• ~1% of lines needed removal
• Saved as .csv file after cleanup

Data Analysis - Jupyter Notebook

• Python was used for analysis with Jupyter

Notebook

• .csv files read into pandas dataframe
• Due to large quantity of data, it was simplified

for speed

• Simplified dataset had every 100th value
• This corresponded to a data point every 5

seconds

• Simplified version was checked against full set
• Negligible difference in large-scale trends

Data Analysis - Plotting

• Of the particulate options, the broadest was > 0.3 um
• This was used for the majority of plots
• All particle size bins agreed very well
• Data was scatter plotted
• Variation of +/- 100 particles was standard
• Significant peaks occurred due to spraying
• Other sources caused peaks as well
• Ventilation system
• Entrance into greenhouse room

Data Analysis - Exponential Decay

• Particulate data followed exponential decay following spray event
• A curve was fit to determine the rate of this decay
• Found that half life was 21 minutes
• Note: only aqueous detection occurs
• Particles could still be in air as gas

Conclusions - Devices

• The devices worked to determine the

number of particles in the air of the green house

• Each of the devices measured very similar

values for particulate data. Differences in data are most likely due to the placement

• f each device. Although they were placed

in the same greenhouse room, they were placed in slightly different areas of the room

Conclusions - Greenhouse Safety for Reentry

• The data of particulates in the air
• ver time shows that greenhouse

reentry is safe after the allotted time

• It was determined that reentry can

actually be earlier than current guidelines permit

Conclusions - Greenhouse to Industrial Settings

• The results obtained for the particulates and

their presence in the air are assumed to be quite different from industrial settings because

• f the area spraying in larger settings as well

as the time it takes to spray a larger area

• Many pesticides are also actually used
• utside.
• Greenhouses can’t take into account wind

and the temperature and disbursement changes due to it

• Wind could affect a larger area of air as well

as the amount of particles present in one location over time

Questions?